Back in March, AMD announced a partnership with Creative Assembly on the development of Total War: Warhammer. The full fruits of their cooperation will be made apparent in June with the release of an update adding DirectX 12 support to the game, but we don’t have to wait until then to evaluate what improvements we can expect. AMD provided us with an early DirectX 12 benchmark build, and we were excited to see how much of a difference, if any, it would have.
Before diving into the benchmark results, it’s worth discussing an important feature of DirectX 12 called Asynchronous Compute. What it basically does is allow graphic and compute tasks to be managed simultaneously. That wasn’t possible in previous versions of DirectX. Workloads had to be prioritized, either programmed by the developers or through automated methods, and wait their turn before they could be completed. In short, our graphics cards weren’t being used as efficiently as they could be. DirectX 12 changes that, and proper support may very well give AMD an edge in performance thanks to the company’s strong commitment to Asynchronous Computing at the hardware level.
The DirectX 12 benchmark loads into a battle between the Empire and the Greenskins. The camera follows a marching ork horde as they move through a ravine flanked by ruined statues and waterfalls of molten lava. It zooms in behind a group of melee units while catapult fire screams overhead. The benchmark soon closes with a greater view of both armies just as their soldiers meet and aerial units tangle in the skies above. Altogether it lasted around 38 seconds.
Unfortunately, the DirectX 11 build we ran did not include the same benchmark. We were thus not able to do a direct 1:1 comparison. Instead, the DirectX 11 numbers came from several custom battles. We attempted to emulate similar conditions and camera angles. Furthermore, because the tests were done on pre-release versions of the DirectX 12 executable and the game itself, we cannot guarantee the same results with the public release of either.
Tests were run on a desktop computer equipped with an Intel i7-3770k, 16GB of DDR3 memory, and an AMD R9 290. Three resolutions were benchmarked – 1920×1080, 2560×1440, and 3440×1440 (for the UltraWide fans out there) – and repeated for accuracy. The game was set to the Ultra preset. The only options it left disabled were depth of field, screen space reflections, and vertical sync.
The DirectX 11 build averaged about 45 frames per second on the R9 290 at 1080p. The framerate dropped to 38 to 42 when the camera was taken close to the ground to inspect a group of soldiers and reached a maximum of 51 when pulled out. Meanwhile, the DirectX 12 benchmark averaged 74.8 frames per second. That’s a stunning 66.22% increase. The minimum for the DX12 benchmark was 68.5 frames with a maximum of 81, 30 frames higher than what we were getting in the DX11 custom battles.
The DirectX 11 build saw an average of 32 frames per second on the R9 290 at 1440p. When zoomed out, the framerate ranged between 32 and 37 frames per second. Zooming in to the thick of battle brought it down to the mid to high 20s, with a minimum of around 26 frames per second. Not surprisingly, the DirectX 12 benchmark once again presented a substantial boost to performance. It gave an average of 52.5 frames per second, a 64.06% increase. The minimum hovered at 48 and the maximum reached 54 frames per second.
The 3440×1440 resolution is a spectacular aspect ratio for video games, especially for the 4X/strategy genre. The extra width allows for a broader view of the battlefield and campaign map. There’s something special about having your empire encompass fill your vision. However, all those extra pixels are much harder to drive. The R9 290 struggled to keep up during the DirectX 11 battles, averaging around 25 frames per second. The minimum dropped to 17 frames per second when the camera was moved closer to the action. A maximum of 32 was only possible by bringing it well above the clashes below. The DirectX 12 benchmark was quite a bit more stable. Its average was 41.0 frames per second, a 64% increase, and only varied slightly from that for the minimum and maximum values (39 and 42 respectively).
Across three resolutions, the DirectX 12 benchmark showed an average improvement of 64.76% compared to the DirectX 11 custom battles. Those are incredible numbers. If they hold true come the release of the DirectX 12 patch, AMD will be giving people an even better reason to take a hard look at the company’s upcoming Polaris and Vega graphic cards.